Saturday, March 30, 2013

Improving the NCAA Basketball Tournament - Earning/Losing At-Large Bids

I've said this before, but Wichita State's run to the Final Four brings the issue into sharp focus - so I'll throw the idea out for consideration.


Every year, several conferences get an outrageous number of at-large bids; every year, we see those teams fail to play to the level at which they were seeded.  It's time for those kind of choke jobs to carry tangible consequences for the conferences.  Let's not forget that MILLIONS of dollars are riding on at-large bids, because tournament revenues are funneled through the conferences, NOT the individual schools.

So, let's take a look at the two conferences represented by Wichita State and Ohio State.  For purposes of this experiment, let's say that a "bad loss" is a loss to a team seeded more than two spots lower; that way, #8/#9 matchups and "First Four" (blech) games are "pick 'em" situations that don't hurt the, who took (or delivered) "bad losses" this year?

The Missouri Valley Conference received two bids; Creighton was an #7 seed, and Wichita State was a #9 seed.  As the tournament progressed, Creighton played up to its seed (they beat #10 Cincinnati and lost to #2 Duke), so that's an "expected result."  The Shockers, of course, played far above their seeding, knocking off #8 Pitt, #1 Gonzaga, #13 LaSalle and #2 Ohio State.  It's now impossible for Wichita State to suffer a "bad loss," since they'll face either #1 Louisville or #2 Duke in the national semifinal.  The MVC overperformed by seeding, delivering two "bad losses," and should be rewarded with an additional at-large bid for next year's tournament.

The Big Ten, on the other hand, has not fared quite so well.  The conference received 7 bids, but #1 Indiana took a "bad loss" to Syracuse, #2 Ohio State took a "bad loss" to Wichita State, #3 Michigan State played to its seeding, #4 Michigan overperformed by one game, #5 Wisconsin took a "bad loss" to Ole Miss, #7 Illinois played to its seed, and #11 Minnesota overperformed by one game.  The conference delivered 2 "bad losses", but underperformed by seeding and suffered 3 "bad losses" itself; the Big Ten should lose at least one at-large bid for next year's tournament.

A quick look at a few other conferences...the Sun Belt performed to its seeding with 2 teams, so they get status quo next year....the SEC overperformed by seeding and administered one "bad loss" (#13 Ole Miss over #5 Wisconsin), so the SEC gets an additional at-large in 2014...the Big East had 8 bids but underperformed, delivering 1 "bad loss" while taking 2 "bad losses" (Georgetown and Notre Dame), so they lose at least one at-large bid next year...the Moutain West's 5 bids resulted in underperformance by seeding and 3 "bad losses," so they lose at least one at-large...Harvard's overperformance and delivery of a "bad loss" brings the Ivy League a second at-large bid next can see how this plays out.

This would redirect a substantial amount of money to conferences who perform well across ALL their representatives in the tournament, give a boost to smaller conferences through the reallocation of at-large bids from underperforming "big" conferences, AND pressure the NCAA to make its seeding process more realistic.  How is this NOT a win-win-win for college basketball?

2013 NCAA Tournament Bracket - March Madness Tournament Brackets - ESPN

Sunday, March 17, 2013

Twitter on Linux - Goodbye Tweetdeck, Hello Polly!

Twitter's impending retirement of its original (v1) API will be the death knell for more than a few third-party clients, one of which is Tweetdeck's original Adobe AIR client.  Sure, Tweetdeck stopped supporting its Adobe AIR clients some time ago, but I and many other Linux users had been quite happy with the Linux variant.  Needless to say, I was not looking forward to searching for a replacement Twitter client.   Ubuntu Linux ships with the Gwibber client, but--to be honest--it wasn't impressive at ALL; the Ubuntu release is actually a downgrade from the development release, because it removes several features that "don't fit" with Ubuntu's Unity GUI.  I spent a bit of time with several decent alternatives, and two of them deserve mention; either Turpial or Hotot would serve as a good 'entry-level' client for a basic Twitter user.  However, I wanted multiple columns, multiple accounts and support for Twitter lists; neither of those clients really fit the bill.  Then, thankfully...I discovered Polly.

Polly is the brainchild of a Brazilian developer by the name of Marcelo Hashimoto (@conscioususer).  It's a straightforward client, written in Python, and I'm EXTREMELY impressed with its features and functionality.  In 15 minutes of playing around with Polly, I had configured its display into a near-clone of the original Tweetdeck UI - separate columns for timeline, mentions, DMs, and the lists I follow (including my own @wesmorgan/bluegrass-area compilation of Central Kentucky sources).  Polly is responsive, low-impact in terms of system resources, and "just works."  Now, the Launchpad page for Polly says that it's still "pre-alpha" code, but it most certainly rises above that label.  Development is ongoing--font/pointsize controls are slated for the next release--and I'm looking forward to watching Polly improve further.

If you're looking for a solid Linux Twitter client, take a LONG look at Polly.  I think you'll be glad that you did.

Tuesday, March 12, 2013

The Tech Side of #socbiz - IBM Support Assistant

One of the most frustrating aspects of systems administration/management is the all-too-frequent discovery that a problem you're experiencing is actually a "known issue", and that you've gone through the hassle of opening a trouble ticket and collecting/transferring a few hundred megabytes of logfiles and debug traces for something that already has a known fix or workaround.  Well, I'm going to help you avoid that - or, at least, make the data collection as easy as possible - where many of IBM's products are concerned...and the tools are FREE!

IBM Support Assistant
(ISA) was developed to speed both the collection and analysis of diagnostic data.  There are two main "pieces" in the ISA architecture: the ISA Lite Data Collectors, which automate the collection of data for specific products, and the ISA Workbench, which lets you drill down into collected data for problem discovery, trend analysis, and the like.   ISA Lite Data Collectors are available for products in every area of IBM's software portfolio, including IBM Connections 3.x/4.0, IBM Sametime, IBM Websphere Application Server, IBM Tivoli Directory Server, IBM DB2, and more; check the ISA webpage for a complete listing.

The other side of the ISA architecture--the ISA Workbench (ISAW)--is where the rubber hits the road.  There are dozens of analysis tools available within the ISAW framework, allowing you to get down to the nitty-gritty of your data quickly and easily.  I've had particularly good results with several tools:
  • Log Analyzer: This Tivoli plugin not only performs basic log analyses, but also allows you to import "symptom catalogs" for various IBM products.  For instance, you can import a symptom catalog for Websphere Application Server 7.x that is based on the current Technote library; this will allow you to catch "known issues" through a log review - possibly doing so BEFORE they become a major problem.
  • Memory Analyzer for Java: This tool is a Java heap analyzer that will help you find memory leaks and manage memory consumption of Java apps.
There are also some promising new tools in the "Tech Preview" category; I'm particularly interested in the "IBM Trace and Request Analyzer for Websphere Application Server," which works to detect delays/hangs in Websphere and HTTP plugin trace files.

I think it's well worth your time to download ISA, collect some sample data and put it through its paces.  ISA is available for both Windows and Linux.  While RHEL and SUSE are the only officially supported Linux variants, I have installed ISAW under Ubuntu 12.10 with no problems noted in my preliminary test cases.  (You'll need to use alien to convert the downloaded .rpm to a .deb file for use with Ubuntu...)

Give it a shot; there's no telling what you might find.